Aurothiomalate Inhibits Transformed Growth by Targeting the PB1 Domain of Protein Kinase Cι*

We recently identified the gold compound aurothiomalate (ATM) as a potent inhibitor of the Phox and Bem1p (PB1)-PB1 domain interaction between protein kinase C (PKC) ι and the adaptor molecule Par6. ATM also blocks oncogenic PKCι signaling and the transformed growth of human lung cancer cells. Here we demonstrate that ATM is a highly selective inhibitor of PB1-PB1 domain interactions between PKCι and the two adaptors Par6 and p62. ATM has no appreciable inhibitory effect on other PB1-PB1 domain interactions, including p62-p62, p62-NBR1, and MEKK3-MEK5 interactions. ATM can form thio-gold adducts with cysteine residues on target proteins. Interestingly, PKCι (and PKCζ) contains a unique cysteine residue, Cys-69, within its PB1 domain that is not present in other PB1 domain containing proteins. Cys-69 resides within the OPR, PC, and AID motif of PKCι at the binding interface between PKCι and Par6 where it interacts with Arg-28 on Par6. Molecular modeling predicts formation of a cysteinyl-aurothiomalate adduct at Cys-69 that protrudes into the binding cleft normally occupied by Par6, providing a plausible structural explanation for ATM inhibition. Mutation of Cys-69 of PKCι to isoleucine or valine, residues frequently found at this position in other PB1 domains, has little or no effect on the affinity of PKCι for Par6 but confers resistance to ATM-mediated inhibition of Par6 binding. Expression of the PKCι C69I mutant in human non-small cell lung cancer cells confers resistance to the inhibitory effects of ATM on transformed growth. We conclude that ATM inhibits cellular transformation by selectively targeting Cys-69 within the PB1 domain of PKCι.

We recently demonstrated that PKC 2 is an oncogene in human lung cancer (1). PKC expression is elevated in non-small cell lung cancer (NSCLC) cell lines and primary tumors, and PKC expression predicts poor clinical outcome (1). PKC expression is driven in a significant subset of NSCLC tumors by tumor-specific amplification of the PKC gene (PRKCI) that resides on chromosome 3q26 (1), the most frequent site of tumor-specific amplification in NSCLC cancers (2,3). PKC stimulates a pro-oncogenic PKC/Par6 3 Rac1 3 Mek1/2 3 Erk1/2 signaling pathway that is required for transformed growth of NSCLC cells in vitro and tumor formation in vivo (4). Expression of either a kinase-deficient PKC mutant (kdPKC) or the Phox and Bem1p (PB1) domain of PKC inhibits oncogenic PKC signaling and blocks transformed growth, indicating the involvement of both PKC kinase activity and PB1 domain interactions in oncogenic PKC signaling (4). PKCmediated transformation requires activation of the downstream effector, Rac1, which is coupled to PKC through PB1-PB1 domain interactions between PKC and the adaptor molecule Par6 (4). Our studies provided the first compelling evidence that PKC is an oncogene in human cancer (1). Subsequent reports have indicated that PKC serves as both a prognostic marker (5,6) and an oncogene in ovarian cancer (6). However, it remains to be determined whether PKC mediates transformed growth of ovarian cancer cells by a similar signaling mechanism.
Based on our results, we reasoned that a small molecule chemical inhibitor of the PKC-Par6 interaction would have potential as a chemotherapeutic agent in the treatment of NSCLC. Therefore, we designed and implemented a novel high throughput drug screening assay based on fluorescence resonance energy transfer to identify compounds that can disrupt the PB1-PB1 domain-mediated binding of PKC to Par6 (7). We used our assay to screen a commercial library consisting of a diverse set of chemical compounds approved for human use. From our screen, we identified several known drugs that potently inhibit PKC-Par6 interactions in vitro (7). Two of the most potent inhibitors identified were the gold compounds aurothiomalate (ATM) and aurothioglucose (ATG). These compounds inhibit the PB1-PB1 domain interaction between PKC and Par6 in a dose-dependent fashion and exhibit potent anti-tumor effects in vitro and in vivo in pre-clinical models of NSCLC (7). Because ATM is currently approved for clinical use in the treatment of rheumatoid arthritis, this compound is a particularly attractive candidate for clinical development as a novel, mechanism-based anti-tumor agent and is the focus of this study.
The PB1 domain is a highly conserved protein-protein interaction domain that is present on a family of signaling proteins. Unique interaction codes exist that specify the ability of PB1 domains to interact with each other (8). Therefore, we wished to determine whether ATM exhibits selectivity toward specific PB1 domain interactions or is a broad specificity inhibitor of PB1-PB1 domain interactions. We also sought to elucidate the molecular mechanism of ATM-mediated inhibition of PB1-PB1 domain interactions and transformed growth. Here we demonstrate that ATM exhibits high selectivity for PB1-PB1 domain interactions involving PKC. Furthermore, by using a combination of sequence alignments, the crystallographic structure of the PKC-Par6 complex, and site-directed mutagenesis, we identify the molecular mechanism by which ATM selectively inhibits the PB1 domain of PKC. Finally, we demonstrate that ATM inhibits transformed growth by selectively targeting the PB1 domain of PKC.
GST Pulldown Assays of PB1-PB1 Domain Interactions-GST fusion proteins were produced in E. coli strain BL21(DE3)pLysS (Novagen) and purified from cell extracts using glutathione-Sepharose 4 Fast Flow beads (Amersham Biosciences). 35 S-Labeled Myc-tagged proteins were co-transcribed/translated in vitro using the TNT T7 coupled reticulocyte lysate system (Promega). For each GST pulldown experiment, 5-10 l of in vitro translated Myc-tagged protein was diluted in 100 l of NETN buffer (20 mM Tris-HCl (pH 8.0), 100 mM NaCl, 0.5% Nonidet P-40, 6 mM EDTA, 6 mM EGTA) containing 1 tablet per 10 ml of Complete Mini, EDTA-free protease inhibitor mixture (Roche Diagnostics). The diluted in vitro translated proteins were preincubated with empty glutathione-Sepharose 4 Fast Flow beads and the indicated amount of ATM for 30 min on a rotating wheel at 4°C. Glutathione-Sepharose beads with immobilized GST fusion protein were added to the supernatant, and the samples were incubated for 60 min on a rotating wheel at 4°C. The beads were washed five times with NETN buffer containing the indicated amount of ATM and boiled in 30 l of Laemmli sample buffer. The samples were resolved in a 10% SDS-polyacrylamide gel, and the gel was Coomassie-stained and vacuum-dried. 35 S-Labeled proteins were detected on a bio-imaging analyzer (BAS-5000, Fujifilm). Quantification of gel band intensities was done using the program Image Gauge (Fujifilm).
Immunoprecipitation of in Vitro Translated p62-pDestHA-p62 or pDestHA-p62R21A (500 ng) was transcribed/translated in vitro in a total volume of 25 l using the TNT T7 coupled reticulocyte lysate system according to the manufacturer's pro-tocol (Promega). Before transcription/translation in vitro, the indicated amount of ATM was added to the reticulocyte lysate. Expression of p62 from these vectors results in two products, HA-tagged p62 and untagged p62. After transcription/translation in vitro, 20 l of the extract was diluted in 200 l of ice-cold NETN buffer (with protease inhibitor mixture and the indicated amount of ATM). The samples were preincubated with a slurry of CL-4B Protein A-Sepharose beads in NETN buffer for 20 min at 4°C on a rotating wheel and then incubated with 0.1 g of anti-HA monoclonal antibody (12CA5) for 1 h and for another 30 min in the presence of bovine serum albumin-saturated Protein A-Sepharose beads in NETN buffer. The beads were washed five times in NETN containing the indicated amounts of ATM or ATG. The beads were resuspended in 15 l of 2ϫ SDS-polyacrylamide gel load buffer and boiled for 5 min. The samples were resolved by SDS-PAGE, and 35 S-labeled proteins were detected on a bio-imaging analyzer (BAS-5000, Fujifilm).
PKC-Par6 Binding Assay-Recombinant biotinylated human Par6 protein was covalently bounded to streptavidincoated 96-well microtiter plates (Nunc, Roskilde, Denmark) by incubation at room temperature for 2 h as described previously (7). Plates were washed twice with PBS/Tween 20 and then once with PBS. After washing, purified PKC/YFP was added to each well along with ATM to achieve the desired final concentration as indicated in the figures, and binding was determined as described previously (7). Wells were then washed with PBS two times, and bound PKC/YFP was determined by measuring fluorescence intensity using a Typhoon 9410 imager (GE Healthcare). Binding analysis of wild-type and C69I and C69V PKC mutants was performed in the presence of increasing amounts of PKC protein as indicated in the figures. Scatchard analysis was used to calculate apparent K d values for binding to Par6.
Cell Transfections, PKC Kinase Assay, and Transformation Assays-Human H1703 NSCLC and HEK 293 cells were obtained from American Type Culture Collection and maintained in adherent culture as described previously (4). H1703 cells were stably infected with pBabe retroviruses containing either FLAG-tagged wild-type human PKC, a human PKC C69I mutant, or empty pBabe virus using puromycin selection as described previously (1,11). Cell lysates were analyzed by immunoblot analysis for FLAG-tagged transgenic PKC, endogenous PKC, and actin as described previously (4). Lysates from HEK 293 cells transfected with wild-type PKC, PKC C69I, or kinase-deficient PKC (kdPKC) were subjected to immunoprecipitation kinase assay using anti-FLAG antibody as described previously (11,12). H1703 cell transfectants were plated in soft agar in the absence or presence of ATM at the concentrations indicated in the figure legends and assessed for soft agar colony formation as described previously (1,4). Plates were analyzed at each ATM concentration, and results are expressed as the mean colony number Ϯ S.E. (n ϭ 5).
Molecular Modeling-Molecular modeling studies were conducted on a Linux work station with the software package Gold (13). The initial three-dimensional structures were obtained using SYBYL version 7.0 (Tripos Inc.) and energy-minimized by using the MAXIMIN routine of SYBYL.

ATM Is a Highly Selective Inhibitor of PB1-PB1 Domain
Interactions Involving PKC-We recently identified the gold compounds ATM and ATG as inhibitors of the PB1-PB1 domain interaction between PKC and the adaptor molecule Par6 (7). Our results demonstrated that not only did these compounds inhibit PKC-Par6 interactions, but they also inhibited PKC signaling to Rac1, a downstream effector of the PKC-Par6 complex and blocked transformed growth of NSCLC cells in vitro and tumorigenicity in vivo (7). Our results suggested that the inhibitory effects of ATM on transformed growth are mediated through disruption of PB1-PB1 domain interactions. However, it remained to be determined whether the growth inhibitory effects of ATM were mediated exclusively through inhibition of PKC-dependent PB1-PB1 domain interactions or whether ATM inhibits other PB1-PB1 domain interactions that might contribute to its cellular effects. To address this question, we developed highly specific pulldown assays for major PB1-PB1 domain interactions using GST fusion proteins and 35 S-labeled binding partners essentially as described previously (8). ATM potently inhibits PKC binding to Par6 (7); therefore, we first assessed whether it would also inhibit PKC binding to p62 (Fig. 1). For this purpose, we used GST pulldown assays to analyze binding between 35 S-labeled human PKC and GST-tagged p62. Because p62 can polymerize via PB1-PB1 domain interactions involving both the OPCA and basic cluster motifs within its PB1 domain (8,14), we utilized a mutant p62 in which Asp-69 within the OPCA motif was mutated to alanine (p62D69A). The D69A mutation makes p62 incapable of forming homopolymers or homodimers but preserves its ability to bind to PKC via its basic cluster motif (8). In the absence of ATM, PKC binds GST-p62 and is recovered in the GST pulldown (Fig. 1A). However, in the presence of ATM, significant inhibition of PKC binding is observed (Fig. 1, A and  B). The mouse homolog of PKC, PKC, also binds p62 via the conserved OPCA motif. Binding of PKC to p62 is similarly inhibited in the presence of ATM (Fig. 1C).
Because ATM inhibits binding of PKC to Par6 (7), our present results suggest that ATM may target either the OPCA domain of PKC or the basic cluster domains of both Par6 and p62. To distinguish between these two possibilities, we next assessed the ability of ATM to inhibit PB1-PB1 domain interactions involving p62. For this purpose, we developed two separate assays to measure p62 dimerization and polymerization, respectively. To measure p62 dimerization, we used a pulldown assay in which we incubate GST-p62D69A immobilized on glutathione-Sepharose beads with in vitro translated, 35 S-labeled p62R21A. These two mutants are incapable of self-dimerization because of the disruption of the OPCA motif (D69A) and basic cluster motif (R21A), respectively (8). However, when combined they are capable of dimerizing with each other ( Fig.  2A), making this a highly specific assay for p62 dimerization. Interestingly, addition of ATM had no appreciable effect on the ability of these proteins to dimerize (Fig. 2, A and B). Similar results were obtained in a related p62 polymerization assay in which wild-type p62 and HA-tagged wild-type p62 were incubated together followed by pulldown using protein A beads and   35 S-labeled p62R21A to GST-p62D69A immobilized on glutathione-Sepharose beads. The two mutants were used because they are unable to polymerize. The experiments were performed as described in the legend to Fig. 1. B, quantitative analysis of p62 dimerization in the presence of the indicated concentration of ATM (n ϭ ϮS.E.). C, the effect of ATM on polymerization of p62 was analyzed by immunoprecipitation of in vitro translated, 35 S-labeled HA-tagged p62. The fact that 40% of the in vitro translated p62 protein lacks the N-terminal HA tag due to translational initiation from the native ATG codon was exploited in a polymerization assay for p62. Immunoprecipitations were performed using an anti-HA antibody. Immunoprecipitated (IP) and co-precipitated proteins were resolved by SDS-PAGE and detected by autoradiography (upper panel). The in vitro translated, 35 S-labeled proteins used as input are shown in the lower gel panel (20% of the input run on the gel). The lack of polymerization/self-interaction of the p62R21A mutant is shown in the lower panel. The results shown are representative of three independent experiments. SEPTEMBER 22, 2006 • VOLUME 281 • NUMBER 38 an anti-HA antibody (Fig. 2C). In this latter assay, the presence of untagged p62 in the precipitate indicates p62 polymerization. In the presence of an inhibitor, the amount of untagged p62 will decrease relative to the amount of HA-tagged p62 precipitated. However, no demonstrable change in polymerization was seen in the presence of ATM (Fig. 2C). As a control, HAtagged p62R21A and untagged p62R21A mutants were incubated together. As expected, these mutants are incapable of polymerizing, and as a result, no untagged p62 is found in the precipitate (Fig. 2C). Our results indicate that ATM does not significantly inhibit p62-p62 interactions. Thus it is unlikely that ATM targets either the basic cluster or OPCA motif of p62, but rather our data are consistent with an effect of ATM on the OPCA motif of PKC.

ATM Inhibition of Oncogenic PKC
To further probe the selectivity of ATM, we assessed the ability of ATM to inhibit PB1-PB1 domain interactions between p62 and NBR1. In this interaction, p62 uses a basic cluster motif to bind to the OPCA motif of NBR1 (8). Using a pulldown binding assay analogous in design to our p62 dimerization assay, we observed efficient binding and pulldown of the 35 S-labeled NBR1 protein with GST-p62D69A (Fig. 3A). Just as was seen with p62 dimerization, addition of ATM to the binding assay had no demonstrable inhibitory effect on p62-NBR1 binding (Fig. 3, A and B). Finally, we assessed whether ATM could inhibit the binding of yet another PB1-PB1 domain interaction. In this case, we designed a specific pulldown assay to measure the binding of MEKK3 to MEK5 similar to that used in our other binding assays (Fig. 4A). ATM had no inhibitory effect on MEKK3-MEK5 binding (Fig. 4, A and B). Taken together, our results demonstrate that ATM exhibits remarkable selectivity for PB1-PB1 domain interactions involving PKC, suggesting that ATM targets the PB1 domain of PKC. Given these observations, we next investigated the molecular mechanism underlying ATM selectivity.
Potential Role of Cysteine Residues in PKC-Par6 Binding and ATM Action-Despite extensive clinical use in the treatment of rheumatoid arthritis, the mechanism of action of gold compounds such as ATM is largely unknown. ATM exhibits antiinflammatory properties thought to be mediated through NFB inhibition; however, the molecular mechanism and the target(s) of this effect are currently unknown (15)(16)(17). Based on gold chemistry, a proposed mechanism for ATM involves formation of thiogold-cysteine adducts on target proteins. For example, ATM inhibits thioredoxin reductase activity by forming a thiogold adduct with a critical active-site cysteine residue    18,19). Because ATM is reactive with protein sulfhydryl groups, we explored the possible involvement of cysteines in ATM-mediated inhibition of PKC-Par6 binding. We first assessed the effect of the sulfhydryl blocking reagent NEM on PKC-Par6 binding. Pretreatment of PKC with 100 M NEM led to significant inhibition of PKC-Par6 binding comparable with that observed in the presence of ATM (Fig. 5A). These results suggested a possible role for one or more cysteine residues in PKC and/or Par6 in PKC-Par6 binding. To determine whether cysteines are also involved in ATM action, we assessed the effect of the sulfhydryl-reducing agent dithiothreitol on ATM-mediated inhibition of PKC-Par6 binding. Addition of 10 M dithiothreitol to the binding reaction reversed ATMmediated inhibition of PKC-Par6 binding (Fig. 5B). Collectively, these data suggest the involvement of one or more cysteine residues in PKC and/or Par6 in PKC-Par6 binding and in ATM-mediated inhibition.
Cysteine 69 in PKC Is a Critical Determinant of PKC-Par6 Binding and ATM Action-Given the possible involvement of cysteines in PKC-Par6 binding and ATM action, we assessed whether the PB1 domains of PKC and/or Par6 contain any cysteine residues that might participate in PKC-Par6 binding and serve as a target for ATM-mediated inhibition. The published crystal structure of the PKC-Par6 complex reveals two highly conserved motifs within the PB1 domains of PKC and Par6 that mediate binding between these proteins (20). PKC utilizes its OPCA motif to bind Par6, whereas Par6 utilizes its basic cluster motif to interact with the OPCA motif on PKC. The crystal structure of the PKC-Par6 PB1 domain complex reveals the presence of a cysteine residue on PKC, Cys-69, at the binding interface between PKC and Par6 where it interacts with Arg-28 within the basic cluster motif of Par6 (Fig. 5C). Arg-28 has been shown previously by site-directed mutagenesis to be important for PKC-Par6 binding (20); however, the importance of Cys-69 has not been explored.
Interestingly, sequence alignment of 10 known human PB1 domains revealed that PKC (and PKC) uniquely contains a cysteine residue, Cys-69, within the OPCA motif not present in the OPCA motifs of other PB1 domains (Fig. 5D).
To assess the role of Cys-69 in PKC-Par6 binding and ATM action, we generated PKC mutants in which Cys-69 was changed to isoleucine (C69I) or valine (C69V), two amino acids frequently found at this position in other PB1 domains (Fig. 5D). We also made a substitution of Cys-69 to the structurally similar serine (C69S). Each PKC mutant was expressed in bacteria and analyzed for Par6 binding as described previously (7) Circular dichroism revealed no global structural changes in PKC as a result of these amino acid substitutions (data not shown). However, distinct differences in the ability of these mutants to bind Par6 were observed. The C69S mutant was unable to bind Par6 (data not shown). In contrast, wild-type PKC, PKC C69I, and PKC C69V all bound Par6 (Fig. 6, A-C). Scatchard analysis of PKC-Par6 binding revealed that wild-type PKC (Fig.  6A), PKC C69I (Fig. 6B), and PKC C69V (Fig. 6C) exhibited very similar apparent K d values for Par6 in the 200 -300 nM range demonstrating that these amino acid substitutions had no apparent effect on the affinity of PKC for Par6. However, although ATM inhibited binding of wild-type PKC to Par6 with an apparent IC 50 of ϳ1 M as expected (7), little or no inhibition was observed with either PKCC69I or PKCC69V in the presence of up to 100 M ATM (Fig. 6D). These results demonstrate that Cys-69 within the OPCA motif of PKC plays an important role in binding to Par6, that substitution of Cys-69 to conserved residues found at this position in other PB1 domains supports binding of PKC to Par6, and that Cys-69 is a critical target of ATM responsible for mediating the inhibitory effects of ATM on Par6 binding. Interestingly, Cys-69 is conserved in mouse PKC, consistent with the fact that ATM inhibits the interaction between PKC and p62 (Fig. 1C).

ATM Inhibits Transformed Growth by Targeting Cys-69 on PKC-Our results demonstrate that ATM selectively inhibits PB1 domain interactions involving PKC by targeting Cys-69
within the OPCA motif of PKC. Our previous studies demonstrated that expression of either dominant negative, kinase-deficient PKC (kdPKC) or the PB1 domain of PKC blocks transformed growth of human NSCLC cells in vitro (4) and that ATM inhibits the transformed growth of NSCLC cells (7). Because ATM selectively inhibits PB1-PB1 domain interactions involving PKC by targeting Cys-69 in vitro, we wished to assess whether this mechanism accounts for the inhibitory effects of ATM on transformed growth. Thus, we generated NSCLC cells that stably express either FLAG-tagged wild-type PKC or PKCC69I (Fig. 7A). Immunoblot analysis using anti-FLAG antibody confirmed expression of the transfected proteins (Fig. 7A). To assess whether the C69I mutation had any effect on the catalytic activity of PKC, we expressed FLAGtagged wild-type PKC, PKC C69I, or kinase-deficient PKC in HEK 293 cells and subjected cell lysates to immunoprecipitation-kinase assays for PKC as described previously (12). PKCC69I exhibited robust kinase activity with a specific activity comparable to wild-type PKC indicating that this amino acid substitution did not affect PKC kinase activity (Fig. 7B). A kinase-deficient PKC allele (kdPKC) served as a negative con-trol and showed little or no kinase activity as expected (12). We next assessed the sensitivity of NSCLC cells expressing wtPKC, PKCC69I, or empty pBabe to ATM-mediated inhibition of colony formation in soft agar (Fig. 7C). As expected, ATM exhibited potent, dose-dependent inhibition of the transformed growth of cells expressing either wild-type PKC or harboring the control plasmid pBabe with an apparent IC 50 of ϳ2 M (Fig.  7C). Cells expressing PKCC69I exhibited transformed growth comparable with that of wild-type PKC and pBabe cells in the absence of ATM. PKCC69I cells formed 50 Ϯ 6 colonies, wtPKC cells 52 Ϯ 19 colonies, and pBabe cells 58 Ϯ 10 colonies (mean Ϯ S.D., n ϭ 5). However, PKCC69I cells were resistant to the inhibitory effects of ATM on transformed growth (Fig.  7C). Taken together these results demonstrate that the PKCC69I mutant is catalytically competent and supports transformed growth but confers resistance to the inhibitory effects of ATM on transformed growth. We conclude that ATM inhibits transformed growth by targeting Cys-69 within the OPCA motif of the PB1 domain of PKC. These results are consistent with the mechanism of ATM-mediated inhibition of PKC-Par6 interactions elucidated in this report. Molecular docking of ATM onto the crystal structure of the PKC PB1 domain predicts formation of a cysteine-aurothiomalate adduct at Cys-69 that protrudes into the binding cleft between PKC and Par6 that is likely to cause steric hindrance to Par6 binding (Fig. 7D).
We recently demonstrated that PKC is an oncogene in NSCLC and elucidated an oncogenic PKC signaling pathway that is required for transformed growth (1,4). PKC activity is required for transformation, because expression of a kinasedeficient, dominant negative PKC allele in NSCLC cells blocks transformed growth and tumorigenicity (1,4). Expression of the PB1 domain of PKC also blocks transformation, indicating the importance of this region of PKC in oncogenic signaling (4). Furthermore, dissection of oncogenic PKC signaling revealed that PKC-dependent transformation requires activation of the small molecular weight GTPase Rac1, which in turn activates Mek1/2-Erk1/2 signaling (4) that is required for transformed growth.
Based on these observations, we concluded that PKC is a critical oncogene in NSCLC required for transformed growth and that PKC is an attractive target for therapeutic interven-tion. Therefore, we devised a high throughput drug screen to identify compounds that can inhibit the PB1-PB1 domain interaction between PKC and Par6, which links PKC to Rac1 (7). We discovered the gold compounds ATM and ATG as potent inhibitors of PKC-Par6 binding and showed that these compounds exhibit potent anti-tumor activity both in vitro and in vivo (7). The anti-tumor activity of these compounds appeared to be dependent upon their ability to inhibit PKC-Par6 interactions, because expression of constitutively active Rac1 in NSCLC cells renders cells resistant to their effects (7). However, it remained possible that the anti-tumor effects of ATM could be mediated by inhibition of other PB1-PB1 domain interactions, or possibly through PB1 domain-independent mechanisms.
Here we have explored the specificity of the inhibitory effects of ATM on PB1-PB1 domain interactions. Our results demonstrate that ATM is a highly selective inhibitor of PB1 domain interactions involving PKC. PKC utilizes the OPCA motif within its PB1 domain to interact with two critical adaptor molecules Par6 and p62. The fact that ATM inhibits both PKC-Par6 and PKC-p62 interactions suggests that ATM targets either the OPCA motif of PKC, which mediates binding of PKC to both Par6 and p62, or the conserved basic cluster motifs of Par6 and p62. The observation that ATM has little or no inhibitory effect on p62 dimerization, p62 polymerization, p62-NBR1 or MEKK2-MEK5 PB1-PB1 domain interactions argues against a mechanism involving the basic cluster motif of PB1 domain-containing proteins, and for a mechanism targeting the OPCA motif on PKC. Our site-directed mutagenesis studies conclusively demonstrate that Cys-69 within the OPCA motif of PKC is a major target of ATM action on PB1-PB1 domain interactions. Mutation of Cys-69 to either isoleucine or valine, two frequently observed amino acids found in this position in other PB1 domain-containing proteins, leads to a mutant PKC that retains binding to Par6 but that is resistant to the inhibitory effects of ATM on Par6 binding. Our data also provide two lines of evidence demonstrating that the inhibitory effects of ATM on transformed growth are also mediated through inhibition of the PB1 domain of PKC. First, the observed IC 50 for ATM-mediated inhibition of transformed growth (ϳ2 M) is in good agreement with the measured IC 50 of ATM for inhibition of PKC-Par6 interactions in vitro (ϳ1 M). Second, expression of a PKC C69I mutant FIGURE 7. ATM blocks transformed growth by targeting the PB1 domain of PKC. A, H1703 NSCLC cells were stably transfected with FLAG-tagged human wild-type (wt) PKC, PKC C69I, or empty vector (pBabe) and analyzed for expression of FLAG-PKC, endogenous PKC, and actin. B, HEK 293 cells were transiently transfected with FLAG-tagged human wild-type PKC, PKCC69I, or kdPKC, and cell lysates were subjected to immunoprecipitation (IP)-kinase assay for PKC activity as described previously (12). C, H1703 NSCLC cells expressing FLAG-tagged wild-type PKC, PKCC69I, or empty vector pBabe were plated in soft agar in the presence of the indicated concentrations of ATM. Soft agar colonies were assessed as described previously (4). Results represent the mean Ϯ S.E. (n ϭ 5). D, molecular model of the PKC-Cys-69-ATM adduct based on the crystal structure of the PKC PB1 domain. The Cys-69-ATM adduct is predicted to protrude into the binding cleft normally occupied by Par6 in the PKC-Par6 complex.